{"corpus":"cpython","kind":"real · dense Python","files":2289,"source_mb":48.3,"index_s":129.95,"parsing_s":120.9,"linking_s":6.98,"symbols":86430,"code_links":339827,"search_fts_p50_ms":0.028,"machine":"Apple M4, local","note":"~37 symbols/file — high symbol density"}
{"corpus":"bench_50k","kind":"real · broad TS/JS","files":49718,"source_mb":456.0,"index_s":839.38,"parsing_s":265.77,"linking_s":566.91,"symbols":108730,"code_links":120770,"search_fts_p50_ms":0.916,"machine":"Apple M4, local","note":"PRE-FIX: pre_tokenize ran the Thai word-segmenter (pythainlp newmm) over multi-MB .d.ts files — 99.8% of linking time"}
{"corpus":"bench_50k","kind":"real · broad TS/JS","files":49718,"source_mb":456.0,"index_s":310.57,"parsing_s":295.57,"linking_s":6.88,"symbols":108730,"code_links":120770,"search_fts_p50_ms":0.931,"db_size_mb":960.6,"peak_rss_mb":1428.0,"machine":"Apple M4, local","note":"POST-FIX: size guard skips newmm on large bodies + C-regex Thai detection. Identical symbol/link counts — no correctness change"}